Morphophonology part #2
More o' moras
Serial moraic adjustment in morphologically complex environments
So I left off on the last post without talking about what happens when more than one mora-adjusting morpheme occurs in sequence. I was considering making this more complicated than I am, with constraints that optimize no clusters over medial clusters over final, et cetera. I have, however, decided to take a simpler route, and make the moraic adjustments in such environments serial, where an affix will take the output form from a previous (less-peripheral) affix's application as a base, and all the rules discussed above still apply. So:
> /akilus-$um/ > akilsum
> /akilsum-$uk/ > akilsumuk (deletion that would be triggered by -$uk blocked by a complex cluster)
> /akilsumuk-$ar/ > akilsumkar
Note that in the above, each round takes the previous form's output as the input for the attachment of a further suffix. This is what I mean by serial application.
This is actually pretty straightforward (yay!) so moving on.
Morphologically-conditioned deletion blocking?
I'm still deciding whether to include something akin to deletion blocking in Piro (or the blocking of optional deletion in Aymara), but I'm leaning towards no. It's complex enough as it is, I think, but stay tuned in case I change my mind (and I could probably be goaded into it...).
Minimal word processes
When a morpheme that is underlyingly smaller than two vocalic moras occurs in an environment that would be predicted to result in a monomoraic word (generally independently, although there will be suffixes that only surface as mora adjustment without any additional material, and these could feasibly lead to such a result as well), these forms are forced to conform to the vowel-dimoraic minimal word restriction in Arrakum. Depending on whether the base form has an onset consonant, this occurs in two different ways: forms with an onset consonant undergo initial reduplication of that consonant followed by the vowel /i/, and forms without an onset consonant undergo lengthening or diphthongization of their vowels (from /a e i u/ to [ā ai ai au]):
/qet#/ > qiqet (but still qet- in derived forms, as illustrated in the previous post on this stuff)
/su#/ > sisu (same deal as qet)
/el#/ > ail (same deal as qet)
/amb#/ > āmb (and again, same deal as qet)
The first one is, again, borrowed from Aymara (surprise!), which has one morpheme (sa "say") which in certain derived environments would lead to monosyllabic output forms like *stwa "I say," but where in most dialects (though some do just have stwa!) this surfaces as sistwa.
Final underlying gemination
Finally out of moraic stuff! All that follows is simple in comparison. First of all, some morpheme-final consonants, when they occur intervocalically, become geminated (it then makes sense to think of these as being underlyingly geminate, since gemination is not contrastive in non-intervocalic positions):
/rihašš#/ > rihaš
/rihašš-+iš/ > rihāššiš
Simple enough. Moving on...
A number of suffixes cause preceding consonants to nasalize. These are inspired by the behavior of the Malagasy verbal prefix aN- (eg /m-aN-petraka/ > mametraka, and lots of similar stuff), such as in the examples below:
/akilus-Nim/ > akilunim
/arrāk-Nim/ > arrāŋim
/eldig-Nim/ > eldiŋgim
/surradiv-Nim/ > surradimim
If the suffix is stop-initial, then a preceding alveolar consonant just becomes homorganic with that stop, and the other consonants become either m or n depending on the same rules as the list below. Otherwise, what these change to is dependent on the consonant being altered:
/p v/ > m
/b/ > mb
/n t s z ł r l '/ > n
/d t'/ > nd
/ŋ š k x q/ > ŋ
/g k'/ > ŋg
no preceding consonant: n
If the final consonant being nasalized is underlyingly geminate, gemination is maintained, unless it becomes a nasal+stop sequence. Final consonant clusters also become geminate, depending on the second element of the cluster:
/rihašš-Nim/ > rihaŋŋim
/valadd-Nim/ > valandim
/qerask-Nim/ > qeraŋŋim
Spirantizing suffixes cause stem-final non-nasals to become fricatives:
mebennud-Sārr > mebennuzār
arrāk-Sārr > arrāšār
/b/ > v
/t, t'/ > s
/d/ > z
/l/ > ł
/r/ > š
/k/ > š OR h (depending on the word)
/p, h, g, k' q, ʔ/ > h
no coda: š
These affect clusters and geminates in similar ways to the nasalizing suffixes described above.
Devoicing suffixes, predictably, cause voiced stops to become devoiced. (Surprise!). This is not quite as predictable as would be expected. It also causes /r l/ to be spirantized.
/t'anub-Tsek/ > t'anupsek
The changes caused by these are as follows:
/b/ > p
/d/ > t
/g/ > k OR q (depending on the word)
/r/ > š
/l/ > ł
no coda: š
(I am too lazy to actually diachronics this stuff right now, but with stuff like this I'm at least hopefully giving the illusion that there did occur some consonant changes in the language's history; especially a shift among the voiceless dorsals)
Again, these pattern the same way wrt geminates and clusters as the patterns described above.
Palatalizing suffixes cause the fricatives /ł s x/, as well as /l r/, to become š (ditto with the epenthetic /r/ when there's no coda). There's not much more to them than that; they have no effects on other consonants:
/šebbeh-+Yti/ > šebbēšti
Finally, geminating suffixes cause the final consonant to become geminated, if it isn't a cluster or already geminated. I'm sure you're shocked. These can combine with palatalizing, devoicing, spirantizing and nasalizing suffixes in ways that are probably pretty predictable.
/arrāk-:$&kāsti/ > ārrakkāsti
All of these can combine with the mora adjustment processes described earlier.
Annnnd I think that's all I have for now. I'll resume working on making the script presentable.